|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 0.9148293843198243, |
|
"global_step": 4500, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 4.872915819438796e-06, |
|
"loss": 3.1058, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 4.745831638877592e-06, |
|
"loss": 3.0496, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 4.618747458316389e-06, |
|
"loss": 3.0295, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 4.491663277755185e-06, |
|
"loss": 3.0068, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 4.364579097193982e-06, |
|
"loss": 2.9939, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 4.237494916632778e-06, |
|
"loss": 2.9796, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 4.110410736071574e-06, |
|
"loss": 2.977, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 3.983326555510371e-06, |
|
"loss": 2.9616, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 3.856242374949167e-06, |
|
"loss": 2.9615, |
|
"step": 4500 |
|
} |
|
], |
|
"max_steps": 19672, |
|
"num_train_epochs": 4, |
|
"total_flos": 2.7840504397824e+18, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|